skills.rest
33may
33mayCOMMUNITY
Explore all skills by this author →
GitHub

stream-chain

Chains sequential agent prompts into multi-step workflows with streaming context flow.

1|1|Updated Nov 28, 2025
Search Tags:#automation#workflow#multi-agent#pipeline#prompt chaining#claude-flow#stream-chain
One-click install
npx skills add https://github.com/33may/robotics --skill stream-chain-33may
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: stream-chain
Source: https://github.com/33may/robotics/tree/main/humanoid/.claude/skills/stream-chain
Command: npx skills add https://github.com/33may/robotics --skill stream-chain-33may

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Coordinating multi-step AI workflows manually requires copying output from one step into the next, losing context and wasting effort. This Skill automates sequential prompt chaining so each agent step receives the full output of the previous one. ## Core Features & Use Cases - Custom Chains: Run arbitrary prompt sequences where each step builds on prior output, with configurable timeouts, verbose logging, and debug mode. - Predefined Pipelines: Execute built-in workflows for codebase analysis, refactoring, test generation, and performance optimization. - Custom Pipeline Definitions: Declare reusable named pipelines in .claude-flow/config.json with their own prompt lists and timeouts. - Use Case: Run a security audit chain that scans for vulnerabilities, categorizes issues by severity, proposes fixes, and generates security test cases in one command. ## Quick Start Run a stream chain that analyzes my codebase structure, identifies improvement areas, and generates an action plan.

Frequently Asked Questions about stream-chain

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a multi-step prompt chain with claude-flow?▼

Use `claude-flow stream-chain run` followed by at least two quoted prompts. Each prompt becomes a step, and the output of each step flows as context into the next one automatically.

What predefined pipelines are available in stream-chain?▼

Four built-in pipelines exist: analysis, refactor, test, and optimize. Run them with `claude-flow stream-chain pipeline <type>`, optionally adding --timeout, --verbose, or --debug flags.

How do I create a custom reusable pipeline?▼

Define it in `.claude-flow/config.json` under the `streamChain.pipelines` key with a name, description, prompts array, and optional timeout. Then execute it by name with `claude-flow stream-chain pipeline <name>`.

Why does my stream chain step time out?▼

Steps default to a 30-second timeout, which is too short for complex analysis or implementation tasks. Increase it with `--timeout 60` or higher, and use `--debug` to inspect where execution stalls.

What are the limitations of stream-chain context flow?▼

Each step supports up to roughly 100K tokens of context, and throughput is about 2-5 steps per minute depending on complexity. Very long chains may hit context limits, so keep prompts focused and specific.

Related Skills

Explore More Software Engineering Skills→
ruvnetruvnet
70.1k

swarm-advanced

Orchestrates multi-agent swarms for research, development, testing, and analysis workflows.

Community
Advanced
eachlabseachlabs
28

eachlabs-workflows

Create and execute multi-step AI workflows by chaining EachLabs models.

Official
Advanced
d-oitd-oit

agent-coordination

Coordinate multiple agents across software development workflows with parallel, sequential, swarm, and hybrid patterns.

Community
Advanced
bjorkgardbjorkgard

stream-chain

Execute multi-agent pipelines through sequential prompt chaining and predefined workflows.

Community
Advanced